With this, we’ve decided to remove Xorg server and other X servers (except Xwayland) from RHEL 10 and the following releases. Xwayland should be able to handle most X11 clients that won’t immediately be ported to Wayland, and if needed, our customers will be able to stay on RHEL 9 for its full life cycle while resolving the specifics needed for transitioning to a Wayland ecosystem. It’s important to note that “Xorg Server” and “X11” are not synonymous, X11 is a protocol that will continue to be supported through Xwayland, while the Xorg Server is one of the implementations of the X11 protocol.
While we recognize the energy behind some distributions and Fedora spins moving towards a similar future, this decision is limited to RHEL 10—we recognize other Linux distributions have different needs and decision structures, and additionally we are not aware of plans for similar efforts in Fedora, nor are we involved in similar efforts besides sharing our knowledge.
A sensible move, now that X.org is no longer really maintained and considered legacy software by everyone who has the skillset and knowledge to actually maintain it in the first place. I know a number of people are very upset about the move to Wayland, but with nobody left willing to work on X.org because it’s effectively unmaintainable, there’s really no other way to go.
If you really want X.org to continue – perhaps you should channel the energy spent on writing angry online comments towards contributing to X.org. However, with even the most knowledgeable and capable X.org developers no longer wanting to have anything to do with X.org, you’re going to be in for a rough ride.
Thom Holwerda,
I see you’ve been drinking a bit of the coolaid
They’re killing it because they don’t want it to exist any more, but this “effectively unmaintainable” is greatly exaggerated. Don’t overlook how long term stable mature projects often simply continue working without needing any significant maintenance.
So on the one hand we could agree that legacy code can be a liability to those wanting to add new features, but on the other hand mature working code doesn’t suddenly stop working and X11 actually has more features than wayland does, so it’s kind of hard to genuinely make the case that X11 is holding wayland back right now. From your wording it doesn’t seem like you appreciate that mature projects largely offset the need for maintenance thanks to the work already having been done a lot time ago. In this light, your statements comes across as somewhat more accusatory than pragmatic.
I really don’t use X11 because I’m a rabid fan of it. Most of us keeping X11 around have been doing so for purely pragmatic reasons. For example remote desktop on raspberry pi under wayland is broken today and that’s an issue for some users who’ve been using it under X11 for decades. I’ll be happy to move to wayland once it’s issues are ironed out (cross my fingers), but the way the wayland community has been dismissive of user needs and aggressively stomping out the only working solution to date is really an extremely ugly path forward for FOSS. The elitism and lack of respect for pragmatic user needs has been on ongoing challenge for linux in general (even beyond wayland). Linux can do better than this, we should do better than this, I just wish the leaders in charge would. I suppose this may be futile and we’ll continue the tradition of breaking things indefinitely without accountability.
The only Linux desktop market that matters commercially (engineering workstations) is 100% based on X11. It was POSIX, X11 and NFS that attracted CAD vendors and users fleeing from Solaris. It is a niche market but one with a lot of money, very high margins and adoption rates – 100% of CAD software is designed for Linux, often Linux only, and the niche is expanding downwards into cheaper and more widely used tools like Matlab, NoMachine etc.
So far, in Linux desktop Linux equals RHEL but people are getting very annoyed with RedHat/IBM and adoption rates of SLES or Oracle Linux are increasing. It is not even about the pricing, it is all the hoops users have to jump over to comply with Redhat licensing. Centos was a popular choice for virtualization precisely because of the flexibility it offered. Then there are questionable decisions, like moving away from NFS and now Xorg, that undermine foundations of the whole platform. None of this is urgent, we haven’t even fully moved to RHEL 8 yet, but this topic is high on agenda in discussions with vendors. Use of Ubuntu or Rocky Linux is no longer as unthinkable as it was even a few months ago.
As for Xorg, honestly, it doesn’t matter. Most people use other/better X11 implementations for access: NoMachine, VNC or commercial X servers. Very few people access workstations via LAN these days. Workstations have moved to data centers, users are working remotely from thin clients (Windows laptops or Macs). But it is important X11, the protocol, is still available and supported on workstations. Commercial software will stay on X11 for as long as it is available on Linux but users also need tools like X11 DEs, Web browsers, file managers, pdf viewers etc regardless of the machine they happen to start them from. While there is some push in making CAD tools less dependent on network transparency (NFS, X11), if that was to truly succeed Linux desktop would have lost its only commercially successful application but I think it is far more likely we will see a deeper fracture in Linux desktop instead.
“The only Linux desktop market that matters commercially (engineering workstations) is 100% based on X11.”
I cannot tell what you are saying. It appears from the rest of your post that you are saying RHEL is the only desktop that matters ( eg. “in Linux desktop Linux equals RHEL” ). Since RHEL already defaults to Wayland ( now, not in RHEL10 ), I do not understand at all the claim of “100% based on X11”. What is?
I agree that this is a sensible move. It will be nice to get everybody on the same track at some point. A couple of points though:
Xwayland is still supported even in RHEL10 and beyond. Xwayland is really just Xorg configured to talk to Wayland instead of trying to use stuff like libinput and KMS directly. This means that much of the Xorg code is going to be maintained for a while. I would not expect big features but security fixes and the like will effectively continue.
Xorg and Wayland have grown more similar over time and share many components. Both Xorg and Wayland use libinput, libdrm, KMS, and Mesa for example. As those are all used with Wayland, they will also continue to be maintained and developed–including new hardware support.
So, the biggest impact here is really that nobody will be bundling up this stuff into Xorg releases at some point. If somebody wanted to do that, they would still be able to rely on many contributions from Red Hat and others for use in “Xorg”.
This is also a long way off still. Sure RHEL10 will come out in 2025 or so but RHEL8 and RHEL9 still exist. RHEL8 uses Xorg by default and is will be supported by Red Hat until 2029. Red Hat will not only have to fix security issues until then but will also have to address any changes required by the kernel and perhaps even a certain amount of new hardware support. RHEL9 is supported until 2032 and, while Wayland is the default, Xorg is still fully supported as an option. So, Red Hat will have to support that too.
Development will slow down but Xorg will remain viable for those that insist on it for a decade yet.
That said, I am not sure how many will. Red Hat may be supporting it longer than anybody cares. With the momentum that Wayland seems to have at the moment, I expect that almost everyone will have moved to Wayland in the next 5 years ( including people that swear they will not ).
In the end, I think it will be a lack of demand, not a lack of support, that kills Xorg.
Anyone upset over the agonizingly inevitable death of Xorg should always consider the standard generic response… “Patches welcome.” If you’re not part of the solution, you’re part of the problem.
“If you’re not part of the solution, you’re part of the problem.” With all due respect, this is BS. Why do open source developers seem to have such awful opinions of their user base? I see this a lot from Gnome devs working on Fedora. Yes, users aren’t paying money for the software. But they are contributing testing. Which is the entire point of the Fedora project. No users, no need for the project. Otherwise it would be vastly cheaper for Red Hat to just dump the new code directly into RHEL. Users serve an invaluable purpose and ignoring their needs just drives them to other projects. Have a little respect, won’t you? It is public knowledge that there are multiple high level bugs preventing people from using Wayland on a day to day basis.
Frankly I don’t see how Xorg is holding Wayland back. Some of the bugs in Wayland have been languishing there for years. They haven’t been fixed because the devs just don’t care or have been paid to focus on other issues. This won’t change anything, except piss off Xorg users. Wayland won’t just magically get better.
In addition, recommending people do what they know is beyond the average users skill set serves no purpose other than a self-stroke of the ego of the person saying it. Projects don’t benefit when devs act like sarcastic douche and take cheap shots at users. Thankfully not all open source devs have disdain for users and regard them as unworthy of opinions or legitimacy. Enough do that it’s a noteworthy problem however. I’m not suggesting people who donate their time & efforts don’t have the right to apply them as they see fit, just that they don’t have to be an asshole about it.